Increasingly, clinical studies show that service and companion dogs can have a significant positive impact on those with physical and mental disabilities. Unfortunately, there is a finite supply of service dogs and the growth potential of tis supply is limited. The main limitation is the 50-70% attrition rate of dogs bred, raised and traine to be companion or service animals. The high attrition rate makes these animals costly and leads to long waiting lists of those in need. There is a clear need for systematic research that helps identify why some dogs are so successful that then leads to a larger supply of certified dogs. A revolution in our understanding of dog cognition has occurred in the past decade, but little of this new understanding has been applied to real world problems. We propose to combine the resources of the Duke Canine Cognition Center and Canine Companion for Independence to identify cognitive traits that make some dogs more successful service dogs than others. First, to quantify individual variability as it relates to success in training, we wil deploy a battery of cognitive tests to evaluate the abilities of dogs that 1) are certified service dogs 2) failed training and 3) are untrained pet dogs. Second, starting as puppies we will use this cognitive battery longitudinally to test a large cohort of dogs before and after they receive formal training. In studying the cognitive abilities of service dogs we will develop a better understanding of what psychological mechanism(s) successful service dogs rely on or are constrained by when helping humans. We can then use this information to better predict which puppies will be successful service dogs - improving the success of training while increasing the potential number of service dogs available.